Unraveling the nature of platinum group metal catalysts’ performance in NO reduction by CO: difference and relevance

Abstract

A CO-SCR reaction, capable of simultaneously removing NO and CO from vehicle exhaust gases, has garnered significant attention. PGM (platinum group metal) catalysts are still the preferred choice due to their exceptional catalytic efficiency. However, an in-depth understanding of these catalysts’ structure–performance relationship is still lacking. In this study, a series of PGM–CeO₂/Al₂O₃ catalysts were prepared with an identical metal loading. Employing operando infrared spectroscopy, we investigated the evolutions of intermediates and combined relevant characterization techniques with reaction kinetics analysis and DFT simulation calculations to elucidate the nature of the difference in activity. PGM–Al₂O₃ catalysts produced N₂O in the CO + NO reaction, while no significant N₂O generation was observed in PGM–CeO₂ catalysts. The activity of Pt-based catalysts was notably worse than that of other catalysts due to “CO poisoning”. Rh–CeO₂ catalysts exhibited optimal reactivity owing to the atomic dispersion of Rh, while some Rh atoms underwent sintering leading to decreased activity after aging. Although Pd nanoparticles suffered an increase in vertical height during aging, moderate adsorptions of CO and NO made the aged Pd–CeO₂ catalysts perform superior to others. It is evident that metal dispersion does not exclusively dictate catalytic performance; factors such as the local coordination environment of the metal and the adsorption properties of reactants on the metal play equally crucial roles.

---

1. Introduction

CO and NO pollutants, primarily originating from motor vehicle exhaust emissions and the combustion of fossil fuels,^1,2 significantly impact human health and environmental safety,^3–7 and thus mitigating their presence holds paramount importance.⁸ Traditional NH₃-SCR (selective catalytic reduction) technology has been extensively employed for flue gas denitration, but poor catalytic efficiency and ammonia escape issues at medium-low temperatures further restrict its applicability.^9,10 A novel denitration technology, CO-SCR, has garnered considerable attention due to its notable advantages in the simultaneous removal of two harmful gases (NO_x and CO),³ high catalytic efficiency,⁷ and the absence of ammonia escape.^11–14 Despite the development of novel CO-SCR catalysts,¹⁵ platinum group metal (PGM) catalysts remain the preferred choice owing to their exceptional catalytic efficiency.^16–19

Previous studies have proved that controlling the catalyst structure is pivotal for achieving outstanding CO-SCR catalytic performance.^20–22 Konstantin Khivantsev et al.²³ synthesized single-atom Ru₁–CeO₂ catalysts and demonstrated exceptional reactivity at low temperatures, with only 0.1–0.5 wt% of Ru proving sufficient for high activity. Ji et al.²⁴ developed a novel Pt catalyst for the CO-SCR reaction by depositing electrically charged Pt single atoms on CuO, which was immobilized on nanoscale CoAlO sheets. This Pt–CuO/CoAlO catalyst exhibited an ultralow Pt loading of 0.02 wt% and achieved a NO conversion rate of 91%. Furthermore, the local microenvironments surrounding the metal could significantly influence the overall performance of the catalyst.^25,26 Tian et al.²⁷ engineered single-atom Pd₁–CeO₂–TS catalysts with unique local microenvironments by subjecting isolated Pd²⁺ ions to ultrafast shock waves through a thermal shock synthesis method, which resulted in significantly enhanced catalytic activity compared to single-atom Pd₁–CeO₂–AT catalysts from atomic trapping methods. Ji et al.²⁸ anchored Ag atoms onto ordered mesoporous WO₃ (m-WO₃) supports with precise control over Ag loading amounts, which modulated the coordination structure of Ag single atoms. It achieved excellent catalytic performance with 0.3 Ag/m-WO₃ (0.3 wt% Ag) and attained a NO conversion rate of 73% and a N₂ selectivity of 100% while exhibiting good stability in durability tests. Despite significant progress made by previous studies, there remains a dearth of research utilizing operando techniques to investigate intermediate species generation and dynamic evolution of active components during catalytic processes under reaction conditions.

This study synthesized a series of PGM–CeO₂ and PGM–Al₂O₃ samples with an identical PGM loading using the initial impregnation method and subjected the samples to aging at 800 °C to simulate the operational conditions of vehicle exhaust gas catalytic purification. Operando infrared spectroscopy was employed to examine the alterations in intermediate species and catalyst morphology during the CO-SCR reaction, and the structure–performance relationship was illustrated by TEM (transmission electron microscopy), XPS (X-ray photoelectron spectroscopy), CO-DRIFTS (diffuse reflectance infrared Fourier transform spectroscopy) and reaction kinetics analysis. The findings indicate that both the electronic properties of PGM itself and structural changes in the catalyst during sintering significantly influence its CO-SCR catalytic performance. The interaction between the metal and support and the strength of metal adsorption on reactants emerge as pivotal factors governing catalyst activity.

2. Experimental section

2.1. Synthesis of PGM–CeO₂ catalysts

CeO₂ support was prepared by heating Ce(NO₃)₃·6H₂O (99.5%, Macklin) in flowing air at 350 °C for 2 h and then at 500 °C for 3 h. Pt (precursor: H₂PtCl₆·6H₂O, Aladdin), Pd (precursor: Pd(NH₄)₂Cl₄, Macklin) and Rh (precursor: Rh((NH₄)₂Cl₅·H₂O, Macklin) were respectively loaded on CeO₂ by the incipient wetness impregnation (IWI) method, and the loading amounts of Pt, Pd and Rh are all 0.5% (weight fraction). The samples were then dried and calcined at 400 °C with a heating rate of 10 °C min⁻¹ for 4 h in a muffle furnace to obtain different loading metals and denoted respectively as PGM(Pt/Pd/Rh)–CeO₂-400. The fresh catalysts were heated to 800 °C with a heating rate of 10 °C min⁻¹ and baked for 10 h to obtain aged catalysts, denoted as PGM(Pt/Pd/Rh)–CeO₂-800. The fresh and aged Pt–CeO₂ catalysts with 1.0% Pt were obtained using the same impregnating method, which were respectively denoted as 1Pt–CeO₂-400 and 1Pt–CeO₂-800.

2.2. Synthesis of PGM–Al₂O₃ catalysts

The PGM–Al₂O₃ catalysts were synthesized by the same impregnating method as PGM–CeO₂ catalysts, and the support was changed to commercial Al₂O₃ (Aladdin). The fresh and aged Pt–Al₂O₃ catalysts with 1.0% Pt were also prepared and were respectively denoted as 1Pt–Al₂O₃-400 and 1Pt–Al₂O₃-800.

2.3. Catalytic activity and kinetic measurements

In a typical CO + NO reaction test, 50 mg of catalyst mixed with ∼0.75 g of quartz sand was placed between the quartz wool in a U-type quartz tube reactor. The CO + NO reaction light-off tests were conducted at a heating rate of 15 °C min⁻¹. The feed gas consisted of 0.4% CO and 0.4% NO balanced with Ar at a total flow rate of 500 mL min⁻¹. NO concentration was monitored with an HPR-20 R&D online mass spectrometer (Hiden Co. Ltd). Before the test, the catalysts were first reduced at 350 °C in 5% H₂/Ar at a flow rate of 100 mL min⁻¹ for 15 min. To eliminate thermal and diffusion effects, the kinetics of the CO + NO reaction and the reaction order were determined under the condition that the NO conversion rate was less than 20%.

2.4. Catalyst characterization

CO-DRIFTS was performed using a Nicolet iS50 FTIR spectrometer equipped with a Pike in situ reaction chamber with a ZnSe window. In a typical measurement, ∼20 mg of catalyst was placed in a sample holder with quartz wool filled at the bottom. Then, Ar (99.999%) at a flow rate of 200 mL min⁻¹ is introduced into the sample to remove air from the system. The sample was first reduced to 5% H₂/Ar with a flow rate of 100 mL min⁻¹ at 350 °C for 15 min. After cooling to room temperature, Ar flow was purged for 1 min and a spectrum was then recorded as a background. After that, a gas flow of 10% CO/Ar (50 mL min⁻¹) was introduced into the reaction cell for 3 min, then switched to Ar (200 mL min⁻¹) and kept for 1 min, and the spectrum was then recorded.

In operando infrared spectroscopy experiments, a Nicolet iS50 FT-IR spectrometer was coupled with an HPR-20 R&D online mass spectrometer to monitor the components of the outlet flow through a Pike high-temperature reactor cell. The reduction operation and background collection steps were performed in the same method as described above. After background collection, a gas stream of 5% NO/Ar at a flow rate of 50 mL min⁻¹ was introduced into the reactor cell for 3 minutes, and the inlet flow was switched to Ar (flow rate of 200 mL min⁻¹) and maintained for 1 min, during which the NO adsorption spectrum was recorded. Then the test flow (0.8% CO and 0.8% NO balanced with Ar at a total flow rate of 100 mL min⁻¹) was switched and the initial gas concentrations were calibrated using the “QGA professional” software. The sample was then heated at a rate of 15 °C min⁻¹ to the desired temperature, during which the infrared spectra were recorded at specific temperatures.

Transmission electron microscopy (TEM) experiments were performed using a JEM-ARM 200F (JEOL) electron microscope. X-ray photoelectron spectroscopy (XPS) experiments were conducted using a Thermo Fisher Scientific K-Alpha spectrometer with a monochromatic Al Kα (1486.6 eV) X-ray excitation source operated at 12 kV. All the catalyst samples were rapidly transferred to vacuum tubes in a glove box filled with Ar and kept under designed atmosphere conditions. The amount of Pt loadings was measured by inductively coupled plasma (ICP) using an Agilent 7700 spectrometer. CO pulse chemisorption measurements were carried out using a PCA-1200 chemical adsorption analyzer instrument and, the catalysts (50 mg) were pretreated in 5% H₂/Ar at 300 °C for 15 min, followed by purging with Ar for 30 min. After the samples were cooled down to 50 °C, a 10% CO/Ar mixture was injected into the reactor repeatedly until CO adsorption was saturated, and the loop volume was 0.20 mL. The dispersion of Pt was calculated by the amount of CO adsorbed by assuming the CO : PGM adsorption stoichiometry to be 1 : 1.

2.5. Computational details

DFT calculations²⁹ were conducted with the Vienna Ab initio Simulation Package (VASP).^30–33 The exchange and correlation energies were processed using generalized gradient approximation (GGA) with the Perdew–Burke–Ernzerhof (PBE)³⁴ form generated from the projected augmented wave (PAW) method.³⁵ The U value for the Ce 4f state was set as 5.0 eV using the DFT+U method.^36,37 The convergence criterion for structural optimization was the maximum force on each atom less than 0.05 eV Å⁻¹ and the energy cutoff was set as 400 eV. The Grimme-D3 method was used to correct the dispersion force.³⁸

The 2 × 4 c(2 × 2) CeO₂ (111) surface with six atomic layers was cleaved from cubic CeO₂. The vacuum layer between plates was 15 Å and used a 2 × 2 × 1 k-point. During structural optimization, the adsorbates and top three layers were relaxed while the bottom three layers were fixed. The climbing-image nudged elastic band (CI-NEB) combined with the Dimer method was used to search the transition states.^39,40 The frequencies of CO adsorbing structures were calculated within the harmonic approximation,^41,42 and all atoms were fixed except CO and its first neighboring atoms during the relaxations.

3. Results and discussion

3.1. CO-SCR activity and operando infrared test analysis of PGM catalysts

Fig. 1a and e respectively depict the NO + CO reaction light-off profiles of the fresh and aged 0.5PGM–CeO₂ samples after reduction. It can be seen that Rh–CeO₂-400 exhibits the best NO reduction activity among the fresh samples (Fig. 1a) and Pd–CeO₂-800 performs better than other aged catalysts (Fig. 1e). Conversely, Pt–CeO₂ samples always display the worst activities compared with other fresh or aged catalysts. Since the molar amount of loaded Pt metal is about one half of Rh or Pd amounts with the same PGM mass fraction in experiments, we prepared 1Pt–CeO₂ to further ensure the same molar amounts of PGM in actual samples and the new 1Pt–CeO₂ sample still shows the worst activity among fresh and aged samples under the same test conditions (Fig. S1, ESI†). The discussion above indicates that calcination causes different effects on various PGM–CeO₂ catalysts.

Fig. 1 NO reduction light-off profiles and operando infrared test analyses of PGM–CeO₂ catalysts. (a) NO reduction light-off profile of PGM–CeO₂-400. Operando infrared spectra of (b) Pt–CeO₂-400, (c) Pd–CeO₂-400 and (d) Rh–CeO₂-400. (e) NO reduction light-off profile of PGM–CeO₂-800. Operando infrared spectra of (f) Pt–CeO₂-800, (g) Pd–CeO₂-800 and (h) Rh–CeO₂-800. The samples were reduced at 350 °C in 5% H₂/Ar for 15 min before tests.

Operando diffuse reflectance infrared spectroscopy was employed to characterize the dynamic evolution of surface species of PGM–CeO₂ catalysts during the CO-SCR reaction and the infrared spectra at various temperatures are presented in Fig. 1b–d and f–h. It can be found that bridging bidentate nitrate (1606 cm⁻¹), monodentate nitrate (1437/1310 cm⁻¹) and free NO₃⁻ (1180 cm⁻¹) rapidly appear with PGM–CeO₂ samples exposed to a NO atmosphere. Notably, both bridging bidentate nitrate and free NO₃⁻ display limited thermal stability and would gradually transition into monodentate nitrate with an increase in temperature. Additionally, carbonate (1234/1356 cm⁻¹) shows a stronger absorption peak at elevated temperatures and thus exhibits greater thermal stability in comparison to carboxylate (1510 cm⁻¹).⁴³ As seen in Fig. 1b and f, when the Pt–CeO₂ sample is subjected to NO and CO atmospheres, notable characteristic absorption peaks are observed at 2180/2157 cm⁻¹ and 2004/1960 cm⁻¹, which are respectively attributed to physically adsorbed CO and the vibrational signal of CO chemisorption on Pt nanoparticles⁴⁴, and their intensities gradually rise with an increase in temperature (>250 °C) and even would persist after NO complete conversion at 450 °C (Fig. 1a and e). The pronounced infrared adsorption signal of CO on the fresh or aged Pt–CeO₂ catalyst indicates that “CO poisoning” may be the primary factor of the diminished activity for Pt–CeO₂.⁴⁵

Compared to Pt based catalysts, Pd catalysts exhibit no significant signal of CO chemisorption (Fig. 1c and g) and it could imply the effective resistance of “CO poisoning” of Pd metal and explain the reason for superior activity under the same conditions. Furthermore, the aged Pd catalyst displays little stronger CO chemisorption compared to the fresh one (260–340 °C), which is consistent with the superior performance of the aged Pd catalyst. Besides, the change of CO chemisorption also reveals that the Pd morphology may change after aging.

The absorption peaks at 2085 cm⁻¹ and 2024 cm⁻¹ in Fig. 1d and h are respectively attributed to the symmetric stretching vibration and asymmetric stretching vibration of CO on Rh single atoms⁴⁶, which indicates that Rh species predominantly exist in a single atom form for both fresh and aged Rh catalysts. Furthermore, it is observed that the strength of CO adsorption on Rh species falls between those of Pt and Pd, and the CO adsorption signals disappear with the temperature exceeding 260 °C, which illustrates that the Rh catalyst exhibits a strong ability to convert NO by CO.

In order to investigate the influence of supports on the CO-SCR reaction activity of PGM catalysts, PGM–Al₂O₃ catalysts were prepared using the same method and the corresponding light-off profiles for the NO + CO reaction are presented in Fig. 2a and e. It is evident that both Al₂O₃ and CeO₂ supported PGM catalysts exhibit similar activity trends and even when being loaded with an equal molar amount of PGM, both fresh and aged samples of 1Pt–Al₂O₃ exhibit the worst CO-SCR activity compared to other catalysts (Fig. S2, ESI†).

Fig. 2 NO reduction light-off profiles and operando infrared test analyses of PGM–Al₂O₃ catalysts. (a) NO reduction light-off profile of PGM–Al₂O₃-400. Operando infrared spectra of (b) Pt–Al₂O₃-400, (c) Pd–Al₂O₃-400 and (d) Rh–Al₂O₃-400. (e) NO reduction light-off profile of PGM–Al₂O₃-800. Operando infrared spectra of (f) Pt–Al₂O₃-800, (g) Pd–Al₂O₃-800 and (h) Rh–Al₂O₃-800. The samples were reduced at 350 °C in 5% H₂/Ar for 15 min before tests.

The operando infrared spectra of PGM–Al₂O₃ samples at different temperatures are presented in Fig. 2b–d and Fig. 2f–h. It is found that a new absorption peak at 2246 cm⁻¹ attributed to N₂O is only observed on the PGM–Al₂O₃ catalysts.⁴⁴ It indicates that the N₂O byproduct cannot be fully reduced by CO for Al₂O₃-supported catalysts, because nitrate species are known for their instability and would facilely decompose into N₂O at elevated temperatures,⁴⁷ which proves that the signal of monodentate nitrate (1297 cm⁻¹) weakens while N₂O production increases at higher temperatures as shown in Fig. 2b–d and f–h.

3.2. Structure–performance relationship analysis

Fig. 3 presents the room temperature CO-DRIFT spectra of the aforementioned catalyst samples. As depicted in Fig. 3a, 2085 cm⁻¹ and 2063 cm⁻¹ are assigned to CO absorption at well-coordinated (WC) Pt sites (terraces) and under-coordinated (UC) Pt sites (steps, edges, and corners) of Pt nanoparticles, respectively.^48,49 In addition, the Pt–CeO₂-400 sample exhibits a weak adsorption peak at 2109 cm⁻¹, representing CO adsorption on the Pt single atom. This suggests that Pt mainly exists in the form of nanoparticles and a few single atoms in the fresh Pt–CeO₂ catalyst. The CO-DRIFT spectra in a flowing CO atmosphere also prove the existence of a Pt single atom, which only has a weak absorption peak at 2109 cm⁻¹ (Fig. S3a, ESI†).⁵⁰ Notably, the intensity of the absorption peak at 2085 cm⁻¹ in the aged sample is significantly enhanced compared with that of the fresh sample (Fig. 3a), which illustrates an increase of WC Pt sites on larger Pt nanoparticles after aging.⁵¹ Similarly, Pt exclusively exists in the form of nanoparticles for the fresh Pt–Al₂O₃ catalyst (Fig. 3d). Remarkably, no significant CO absorption signals on Pt–Al₂O₃-800 appear, perhaps because Pt nanoparticles are encapsulated by support during the aging process, seriously hindering CO adsorption.^52,53

Fig. 3 Room temperature CO-DRIFT spectra of (a) Pt–CeO₂, (b) Pd–CeO₂, (c) Rh–CeO₂, (d) Pt–Al₂O₃, (e) Pd–Al₂O₃ and (f) Rh–Al₂O₃ catalysts. The samples were reduced at 350 °C in 5% H₂/Ar for 15 min before testing.

The two absorption peaks in the CO-DRIFT spectrum of the Pd–CeO₂ catalysts are respectively attributed to CO linear (2089 cm⁻¹) and bridging (1971 cm⁻¹) adsorption on the Pd surface, indicating the form of Pd nanoparticles.⁵¹ Notably, a slight red-shift is found for the CO linear and bridging adsorption at Pd–CeO₂-800 (Fig. 3b). Two forms of bridging adsorption (1980 cm⁻¹ and 1926 cm⁻¹) occur on Pd–Al₂O₃ and a strong CO adsorption signal could still be observed at the aged Pd–Al₂O₃ catalyst, reflecting that Pd is still well exposed on the surface (Fig. 3e). This explains why the CO-SCR performance of the aged Pd–Al₂O₃ catalyst is significantly better than that of aged Pt–Al₂O₃ catalysts.

According to the two CO adsorption peaks (2084 cm⁻¹ and 2010 cm⁻¹) of the Rh–CeO₂-400 catalyst, it is determined that Rh is dispersed on the surface in the form of single atoms for the fresh samples (Fig. 3c). Compared with Rh–CeO₂-400, Rh–CeO₂-800 exhibits a slight blue-shift (∼9 cm⁻¹) of symmetric stretching vibration (2093 cm⁻¹) and a decreased peak of the asymmetric stretching vibration (2010 cm⁻¹). The CO-DRIFT spectra of Rh–CeO₂-400 and Rh–CeO₂-800 in a flowing CO atmosphere show an obvious adsorption signal of Rh single atoms and a blue-shift phenomenon is also observed (Fig. S3b and c, ESI†).⁵⁴ It may be caused by the different morphologies of Rh species and further be explained in detail based on the simulation calculation below. Fig. 3f indicates that Rh is still distributed as a single atom on the Al₂O₃ surface. No blue shift and weak CO adsorption on Rh–Al₂O₃-800 further suggest that partial Rh single atoms could be encapsulated by the support during aging. The decrease of active sites could cause aged Rh-based catalysts to exhibit weaker activity than aged Pd-based catalysts.

Fig. 4 presents the TEM images of the PGM–CeO₂ catalysts. After reduction treatment, a small amount of Pt single atoms and nanoparticles with an average particle size of 1.2 nm are observed in Fig. 4a and the average size of Pt nanoparticles on Pt–CeO₂-800 increases up to 2.4 nm (Fig. 4d), which indicates that Pt species undergo severe sintering during aging. As shown in Fig. 4b and e, Pd exists in the form of nanoparticles in both fresh and aged samples and their difference in the average particle size is only 0.3 nm (Fig. 4b and e), suggesting no significant sintering of Pd nanoparticles during the aging process. Only Rh single atoms are observed in Rh–CeO₂-400 (Fig. 4c), while Rh single atoms and Rh clusters coexist in Rh–CeO₂-800 (Fig. 4f). The high atomic utilization of single atoms may explain why the fresh Rh–CeO₂ catalysts exhibit better CO-SCR reactivity than the aged samples. Ultimately, the PGM morphologies in Fig. 4 are completely consistent with CO-DRIFTS analysis (Fig. 3).

Fig. 4 TEM images of (a) Pt–CeO₂-400, (b) Pd–CeO₂-400, (c) Rh–CeO₂-400, (d) Pt–CeO₂-800, (e) Pd–CeO₂-800 and (f) Rh–CeO₂-800 after reduction treatments.

Table 1 shows the dispersion test of PGM–CeO₂ catalysts. The dispersion of Pt–CeO₂-800 (52.9%) is significantly lower than that of Pt–CeO₂-400 (76.4%), demonstrating the obvious sintering of Pt during the aging process, which is also confirmed by the TEM results of Pt–CeO₂ catalysts (Fig. 4a and d). The dispersion of Pd–CeO₂-800 (47.8%) is decreased by 34% compared with that of Pd–CeO₂-400 (72.8%), which is different from a small difference in the mean particle size of Pd (Fig. 4b and e). This could be ascribed to Pd nanoparticles growing horizontally during the aging process, resulting in a relative decrease of the exposed Pd active sites. The dispersion of Rh–CeO₂-800 (81.1%) is slightly lower than that of Rh–CeO₂-400 (92.7%), indicating aggregation of a small number of Rh single atoms after aging, which is also consistent with TEM results (Fig. 4f).

Table 1 Main states and dispersion of PGM–CeO₂ catalysts after reduction

Sample	Loading^a (wt%)	Main morphologies^b	Dispersion^c (%)
a Measured by ICP. b Determined by CO-DRIFTS as shown in Fig. 3 and TEM images as shown in Fig. 4. c Determined by CO chemisorption.
Pt–CeO2-400	0.61	Single atom and nanoparticle	76.4
Pt–CeO2-800	0.61	Nanoparticle	52.9
Pd–CeO2-400	0.51	Nanoparticle	72.8
Pd–CeO2-800	0.51	Nanoparticle	47.8
Rh–CeO2-400	0.58	Single atom	92.7
Rh–CeO2-800	0.58	Single atom and cluster	81.1

---

XPS spectra were used to analyze the chemical states of Pt, Pd and Rh elements, and the results are shown in Fig. 5. The proportion of Pt⁰ in Pt–CeO₂-400 is 45.2% (Fig. 5a) and that in Pt–CeO₂-800 increases to 53.5% (Fig. 5d), which indicates that aging causes an increase in size of Pt particles, consistent with TEM analysis (Fig. 4d). In contrast, the proportions of Pd⁰ in Pd–CeO₂-400 and Pd–CeO₂-800 are 22.8% and 41.1%, respectively (Fig. 5b and e). Combined with TEM images, it is inferred that Pd in Pd–CeO₂-400 is nearly single-layer dispersed and the number of Pd nanoparticle layers increases significantly after aging. The proportion of Rh⁰ in Rh–CeO₂-400 is 0% (Fig. 5c) and the Rh⁰ amount increases to about 29.2% (Fig. 5f) in Rh–CeO₂-800, which illustrates that Rh exists in the form of single atoms in Rh–CeO₂-400 and some Rh single atoms aggregate to form clusters during the aging process (Fig. 4f). The XPS spectra of PGM–Al₂O₃ samples are depicted in Fig. S4 (ESI†). In Pt–Al₂O₃-400, Pt⁰ constitutes 41.3% of the total (Fig. S4a, ESI†), while no Pt atoms are discernible in Fig. S4d (ESI†), suggesting substantial encapsulation of Pt atoms by Al₂O₃ during aging. This phenomenon also accounts for the diminished activity of Pt–Al₂O₃-800 compared to that of Pt–Al₂O₃-400 as illustrated in Fig. 2a and e. Conversely, the chemical states of the fresh and aged Pd–Al₂O₃ and Rh–Al₂O₃ do not show significant changes, due to the dispersion of Pd and Rh nanoparticles on Al₂O₃.

Fig. 5 XPS spectra of (a) Pt–CeO₂-400, (b) Pd–CeO₂-400, (c) Rh–CeO₂-400, (d) Pt–CeO₂-800, (e) Pd–CeO₂-800 and (f) Rh–CeO₂-800 after reduction treatments. The samples were reduced at 350 °C in 5% H₂/Ar for 15 min before testing. Only Pt²⁺ and Pd²⁺ were deconvolved to represent their positive charges, as conducted by other reports.⁵⁵

The chemical state of Ce is also analyzed by XPS, and the results are shown in Fig. S5 (ESI†). The proportion of Ce³⁺ is nearly the same (26.0%) in PGM–CeO₂-400 catalysts, and the proportion of Ce³⁺ in PGM–CeO₂-800 catalysts decreased to some degree. The proportion of Ce³⁺ in Pd–CeO₂-800 shows a significant decrease (4.6%) (Fig. S5e, ESI†). The decrease of the Ce³⁺ proportion means a decrease in the number of oxygen vacancies.⁵⁶ However, Pd–CeO₂-800 shows good catalytic activity, which indicates that oxygen vacancies did not have a good promoting effect on the CO-SCR reaction. Instead, the order degree of the CeO₂ lattice is greatly improved at high temperatures, and more lattice oxygen is produced.⁵⁷ The increase in the amount of lattice oxygen can enhance the strong interaction between metal and CeO₂,⁵⁸ which promotes the electron transfer and oxygen migration between metal and support,⁵⁹ thus causing the best performance of Pd–CeO₂ among PGM–CeO₂-800 catalysts.

The analysis of reaction kinetics is presented in Fig. 6 and both fresh and aged Pt–CeO₂ catalysts exhibit the lowest activation energy (E_a), suggesting favorable kinetics to catalyze the CO-SCR reaction (Fig. 6a and b). Besides, the NO reaction order is the highest (Fig. 6c and d), indicating too weak NO adsorption with insufficient residence time to undergo a complete reaction. Considering operando infrared testing results (Fig. 1b and f), it is concluded that CO occupies a majority of active sites and significantly impedes NO adsorption and activation, due to strong CO adsorption. Due to approximately identical molecular weights of NO and CO, accurately measuring the CO reaction process using mass spectrometry detection methods proves challenging; therefore, this paper focuses on the NO reaction order analysis. In summary, “CO poisoning” emerges as a fundamental factor of diminished CO-SCR activity for Pt–CeO₂ catalysts.

Fig. 6 Arrhenius plots of (a) PGM–CeO₂-400 and (b) PGM–CeO₂-800. Reaction orders of (c) PGM–CeO₂-400 and (d) PGM–CeO₂-800. The samples were reduced at 350 °C in 5% H₂/Ar for 15 min before testing.

The E_a value of Rh–CeO₂ is lower than that of Pd–CeO₂ in fresh samples, whereas it is opposite in aged samples (Fig. 6a and b), which is consistent with the CO-SCR activity trend depicted in Fig. 1. The reaction order of Pd–CeO₂-400 (0.22) is significantly lower than that of other samples, indicating that strong NO adsorption could partially inhibit CO adsorption. In contrast, both NO reaction orders of the fresh and aged Rh–CeO₂ samples are moderate, which is favorable for the NO reaction with CO. Based on that, due to the high activation energy and low NO reaction order, the CO-SCR activity of fresh Pd–CeO₂ is inferior to that of Rh–CeO₂.^60,61 Conversely, the reaction orders of Pd–CeO₂ and Rh–CeO₂ in aged samples are moderate, while the activation energy of Pd–CeO₂-800 is lower than that of Rh–CeO₂-800, which further demonstrates the superior catalytic performance of Pd–CeO₂-800.

3.3. DFT calculation studies

Based on the analysis of TEM images and dispersion test results of fresh and aged Pd–CeO₂ catalysts, we respectively constructed monolayer Pd₆ and bilayer Pd₁₃ structures to mimic the morphology of Pd nanoparticles in fresh and aged Pd–CeO₂ catalysts.^62,63 The vibrational frequencies of CO linear and bridging adsorption on the two structures were calculated and the results are shown in Fig. 7a–d. Notably, the calculated vibrational frequency is always smaller than the experimental value.⁶⁴ It is evident that CO vibrational frequencies of both linear and bridging on Pd₁₃ are lower than those on Pd₆, aligning with the red-shift phenomenon illustrated in Fig. 3b. This further suggests that aged Pd–CeO₂ poses a larger vertical height of Pd nanoparticles than fresh Pd–CeO₂. Similarly, the Rh₃ structure was used to mimic the Rh cluster morphology on the CeO₂ surface and the CO vibrational frequency on the Rh single atom and Rh₃ cluster was calculated. The CO vibrational frequency on the Rh₃ cluster is 10.2 cm⁻¹ higher than that on the Rh single atom (Fig. 7e and f), which is consistent with the blue-shift phenomenon (9 cm⁻¹) shown in Fig. 3c. This indicates that the increase in the CO vibrational frequency of the aged Rh–CeO₂ sample is due to the sintering of the Rh single atom.

Fig. 7 Vibration frequency calculation and the corresponding structure of CO adsorption on Pd₆, Pd₁₃, Rh₁ and Rh₃ on the CeO₂ surface. (a) CO linear and (b) CO bridging adsorption on Pd₆. (c) CO linear and (d) CO bridging adsorption on Pd₁₃. CO linear adsorption on (e) Rh₁ and (f) Rh₃. Color: Ce, yellow; O, red; C, grey; Pd, cyan; and Rh, yellow.

Combined with the previous analysis, we respectively constructed Pt₁₃, 2Rh₁, monolayer Pd₆ and bilayer Pd₁₃ structures to mimic the morphology of the PGM–CeO₂ catalysts. The two reaction processes (preferential CO and preferential NO adsorption path) and the energy barriers are shown in Fig. 8, and the relaxed structure of each intermediate species and transition state structures are presented in Tables S1–S8 (ESI†). As can be seen from the energy barrier diagram, the preferential NO adsorption path (red line) overall exhibits lower energy. The 2Rh₁ structure has a much higher maximum activation energy (E_a,max) on the preferential CO adsorption path (black line) than that on the preferential NO adsorption path, while the E_a,max values of the two paths in Pt₁₃, Pd₆ and Pd₁₃ structures are all not significantly different.

Fig. 8 Energy profile of the CO-SCR reaction pathway on (a) Pt–CeO₂, (b) Rh–CeO₂, (c) Pd₆–CeO₂ and (d) Pd₁₃–CeO₂ structures. The black line represents the preferential CO adsorption path and the red line represents the preferential NO adsorption path.

A preferential CO adsorption path involving the oxygen vacancy (O_v) was also designed using the same strategy. The reaction path and corresponding energy barriers are shown in Fig. S6 (ESI†), and the relaxed structure of each intermediate species and transition state structures are shown in Tables S9–S12 (ESI†). The selective adsorption of NO to oxygen vacancies leads to more migration steps and thus more reaction steps, and the overall reaction path is not significantly preferred in energy compared with the path without the oxygen vacancy. Accordingly, the surface oxygen vacancy has no significant promotion for the CO-SCR activity of the PGM–CeO₂ catalyst.

The oxidation state of metal plays an important role in affecting the absorption/desorption of intermediates.^65,66 Thus we calculated the dynamic change of the Bader charge of Rh for the Rh–CeO₂ catalyst in the above two reaction paths (preferential CO adsorption path and preferential NO adsorption path) as an example, and the results are shown in Fig. S7 (ESI†). It is obvious that the adsorption of CO and NO by the Rh atom and the migration of the O atom would mostly cause the increase of the value of the Bader charge. In particular, the migration of the O atom causes the values to rise to 1.6 (Fig. S7a, ESI†) and 1.43 (Fig. S7b, ESI†), respectively; they decrease significantly after the migration of the N atom and the desorption of CO₂ and N₂. The migration of the N atom results in the value of Bader charge respectively decreasing to 0.91 (Fig. S7a, ESI†) and 0.52 (Fig. S7b, ESI†). It is worth noting that the migration of O atoms on NO will cause a significant increase in the Bader charge of Rh atoms, indicating that Rh atoms are oxidized by migrating O atoms, forming an oxidized Rh species with high catalytic properties. This case suggests that the chemical state of the metal undergoes dynamic evolution during catalysis.

4. Conclusions

In this study, we synthesized a series of fresh and aged PGM–CeO₂/Al₂O₃ catalysts by the initial impregnation method with an identical loading (0.5 wt%). The CO + NO reaction activity of the catalysts was evaluated, revealing that the metal properties, metal form, and metal–support interaction significantly influence the catalytic activity. Operando infrared spectroscopy was employed to investigate the changes in catalytic intermediate species. Additionally, reaction kinetics analysis, relevant characterization techniques, and DFT simulation calculations were combined to elucidate the structure–activity relationship. The findings are as follows:

(I) Operando infrared spectroscopy revealed that PGM–Al₂O₃ catalysts produce N₂O in the CO + NO reaction, while no significant N₂O generation was observed in PGM–CeO₂. The interaction between CeO₂ and PGM is conducive to the adsorption and activation of CO, which can further reduce N₂O to N₂. However, the trends in CO + NO reaction activity for both PGM–CeO₂ and PGM–Al₂O₃ catalysts remain consistent: Rh-based catalysts exhibit superior activity in fresh samples whereas Pd-based catalysts show optimal performance in aged samples; Pt-based catalysts display the worst activity in both fresh and aged samples.

(II) The NO reaction order of PGM–CeO₂ was determined. Both fresh and aged Pt–CeO₂ catalysts exhibited the highest order, suggesting the relatively weak adsorption of NO. This finding is consistent with the phenomenon of strong CO adsorption observed through operando infrared spectroscopy. Consequently, “CO poisoning” emerges as a fundamental factor contributing to poor CO-SCR activity in Pt-based catalysts. Moreover, Pt nanoparticles becoming encapsulated during aging further hinders NO adsorption and activation.

(III) The fresh Pd–CeO₂ exhibits a higher E_a value and a lower NO reaction order compared to Rh–CeO₂, resulting in lower CO-SCR activity. In aged samples, the NO reaction order of Pd–CeO₂ and Rh–CeO₂ is appropriate, but the E_a value of Pd–CeO₂ is lower than that of Rh–CeO₂. Thus, the aged Pd–CeO₂ catalysts displayed more excellent catalytic performance. Due to the increase in the amount of lattice oxygen, the strong interaction between Pd and CeO₂ is enhanced, which promotes the electron transfer and oxygen migration between Pd and CeO₂, enabling the aged Pd–CeO₂ catalysts to display superior catalytic performance.

(IV) Rh-based catalysts are present in single atom form while Pt and Pd-based catalysts exist as nanoparticles. After aging, the number of stacking layers of Pd atoms increases. Combining the catalyst's CO-SCR activity data, it can be clearly concluded that metal dispersion is not the only factor affecting catalyst activity; the E_a value and adsorption strength of the metal to reactants are equally critical.

Data availability

Data will be made available on request.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (22162005) and the Innovation Project of Guangxi University of Science and Technology Graduate Education (GKYC202333). We would also thank Liao from Shiyanjia Lab (https://www.shiyanjia.com) for the XPS tests.

References

1. T. Cai, Z. Teng, Y. Wen, H. Zhang, S. Wang, X. Fu, L. Song, M. Li, J. Lv and Q. Zeng, J. Hazard. Mater., 2022, 440, 129772.
2. T. Tanabe, T. Imai, T. Tokunaga, S. Arai, Y. Yamamoto, S. Ueda, G. V. Ramesh, S. Nagao, H. Hirata, S.-I. Matsumoto, T. Fujita and H. Abe, Chem. Sci., 2017, 8, 3374–3378.
3. A. Patel, P. Shukla, T. E. Rufford, V. Rudolph and Z. Zhu, Chem. Eng. J., 2014, 255, 437–444.
4. T. Baidya, P. Bera, B. D. Mukri, S. K. Parida, O. Kröcher, M. Elsener and M. S. Hegde, J. Catal., 2013, 303, 117–129.
5. S. Xie, W. Tan, Y. Li, L. Ma, S. N. Ehrlich, J. Deng, P. Xu, F. Gao, L. Dong and F. Liu, ACS Catal., 2022, 12, 2441–2453.
6. F. Xing, J. Jeon, T. Toyao, K.-I. Shimizu and S. Furukawa, Chem. Sci., 2019, 10, 8292–8298.
7. T. Li, L. Li, J. Wang, Y. Wu, Y. Wang and M. Li, Inorg. Chem. Commun., 2023, 150, 110018.
8. C. Asokan, Y. Yang, A. Dang, A. B. Getsoian and P. Christopher, ACS Catal., 2020, 10, 5217–5222.
9. K. Guo, J. Ji, W. Song, J. Sun, C. Tang and L. Dong, Appl. Catal., B, 2021, 297, 120388.
10. L. Han, S. Cai, M. Gao, J.-Y. Hasegawa, P. Wang, J. Zhang, L. Shi and D. Zhang, Chem. Rev., 2019, 119, 10916–10976.
11. E. Fernández, L. Liu, M. Boronat, R. Arenal, P. Concepcion and A. Corma, ACS Catal., 2019, 9, 11530–11541.
12. K. Pan, Y. Yao, D. Cui, F. Gao, M. Liu, J. Dan, B. Dai, X. Guo, Y. Zhang and F. Yu, J. Environ. Chem. Eng., 2024, 12, 112498.
13. K. Zhao, Z. Li, T. Yan, Z. Gao, Q. He, Q. Wang, W. Liu, J. Zhang, X. Zhao, L. Liu and C. Liu, J. Environ. Chem. Eng., 2023, 11, 109757.
14. L. Ilieva, G. Pantaleo, N. Velinov, T. Tabakova, P. Petrova, I. Ivanov, G. Avdeev, D. Paneva and A. M. Venezia, Appl. Catal., B, 2015, 174–175, 176–184.
15. M. Lykaki, E. Pachatouridou, S. A. C. Carabineiro, E. Iliopoulou, C. Andriopoulou, N. Kallithrakas-Kontos, S. Boghosian and M. Konsolakis, Appl. Catal., B, 2018, 230, 18–28.
16. L. Qi, M. Babucci, Y. Zhang, A. Lund, L. Liu, J. Li, Y. Chen, A. S. Hoffman, S. R. Bare, Y. Han, B. C. Gates and A. T. Bell, J. Am. Chem. Soc., 2021, 143, 21364–21378.
17. H. Yoshida, K. Koizumi, M. Boero, M. Ehara, S. Misumi, A. Matsumoto, Y. Kuzuhara, T. Sato, J. Ohyama and M. Machida, J. Phys. Chem. C, 2019, 123, 6080–6089.
18. Q. Xu, X. Cheng, N. Zhang, Y. Tu, L. Wu, H. Pan, J. Hu, H. Ding, J. Zhu and Y. Li, Nano Res., 2023, 16, 8882–8892.
19. J. Wang, F. Gao, H. Yi, H. Liu, T. Xiong, Y. Du, Y. Zhou, E. Duan and X. Tang, Sep. Purif. Technol., 2023, 325, 124684.
20. Y. Gao and D. Wang, CCS Chem., 2024, 6, 833–855.
21. S. Chen, Z. Zhang, W. Jiang, S. Zhang, J. Zhu, L. Wang, H. Ou, S. Zaman, L. Tan, P. Zhu, E. Zhang, P. Jiang, Y. Su, D. Wang and Y. Li, J. Am. Chem. Soc., 2022, 144, 12807–12815.
22. Y. He, J. Liu, G. Zhang, Y. Wang, Y. Zhao, G. Li, Y. Zhang and D. Lv, Chem. Eng. J., 2023, 465, 142856.
23. K. Khivantsev, N. R. Jaegers, H. A. Aleksandrov, I. Song, X. I. Pereira-Hernandez, M. H. Engelhard, J. Tian, L. Chen, D. Motta Meira, L. Kovarik, G. N. Vayssilov, Y. Wang and J. Szanyi, J. Am. Chem. Soc., 2023, 145, 5029–5040.
24. Y. Ji, S. Liu, S. Song, W. Xu, L. Li, Y. Zhang, W. Chen, H. Li, J. Jiang, T. Zhu, Z. Li, Z. Zhong, D. Wang, G. Xu and F. Su, ACS Catal., 2022, 13, 224–236.
25. M. Hou, L. Zheng, D. Zhao, X. Tan, W. Feng, J. Fu, T. Wei, M. Cao, J. Zhang and C. Chen, Nat. Commun., 2024, 15, 1342.
26. S. Cheng, F. Xu, S. Yang, B. Zhang, W. Song, X. Zhu, W. Tan, C. Sun and L. Dong, Environ. Sci. Technol., 2024, 58, 8955–8965.
27. J. Tian, R. Kong, Z. Wang, L. Fang, T. He, D. Jiang, H. Peng, T. Sun, Y. Zhu and Y. Wang, ACS Catal., 2023, 14, 183–191.
28. Y. Ji, X. Chen, S. Liu, S. Song, W. Xu, R. Jiang, W. Chen, H. Li, T. Zhu, Z. Li, Z. Zhong, D. Wang, G. Xu and F. Su, ACS Catal., 2023, 13, 1230–1239.
29. W. Kohn, A. D. Becke and R. G. Parr, J. Phys. Chem., 1996, 100, 12974–12980.
30. G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1993, 47, 558–561.
31. G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 49, 14251–14269.
32. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
33. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
34. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
35. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979.
36. Q. Zhou, C. Zhou, Y. Zhou, W. Hong, S. Zou, X.-Q. Gong, J. Liu, L. Xiao and J. Fan, Catal. Sci. Technol., 2019, 9, 2960–2967.
37. H. Y. Kim, H. M. Lee and G. Henkelman, J. Am. Chem. Soc., 2012, 134, 1560–1570.
38. S. Grimme, J. Antony, S. Ehrlich and H. Krieg, J. Chem. Phys., 2010, 132, 154104.
39. G. Henkelman and H. Jonsson, J. Chem. Phys., 1999, 111, 7010–7022.
40. J. Kastner and P. Sherwood, J. Chem. Phys., 2008, 128, 014106.
41. H. V. Thang, G. Pacchioni, L. DeRita and P. Christopher, J. Catal., 2018, 367, 104–114.
42. C. Dong, Z. Gao, Y. Li, M. Peng, M. Wang, Y. Xu, C. Li, M. Xu, Y. Deng, X. Qin, F. Huang, X. Wei, Y.-G. Wang, H. Liu, W. Zhou and D. Ma, Nat. Catal., 2022, 5, 485–493.
43. G. Cheng, Z. Cai, X. Song, X. Chen, W. Dai and X. Fu, Appl. Catal., B, 2022, 304, 120988.
44. F. Gao, Y. Tang, J. Liu, K. Pan, M. Zhou, G. Qian, M. Liu, F. Yu, J. Dan and B. Dai, Chem. Eng. J., 2023, 474, 145713.
45. H. Cheng, J. Xia, M. Wang, C. Wang, R. Gui, X. Cao, T. Zhou, X. Zheng, W. Chu, H. Wu, Y. Xie and C. Wu, J. Am. Chem. Soc., 2022, 144, 22018–22025.
46. D. Wu, S. Liu, M. Zhong, J. Zhao, C. Du, Y. Yang, Y. Sun, J. Lin, S. Wan, S. Wang, J. Huang, Y. Yao, Z. Li and H. Xiong, ACS Catal., 2022, 12, 12253–12267.
47. P. Lu, L. Ye, X. Yan, X. Chen, P. Fang, D. Chen, D. Chen and C. Cen, J. Hazard. Mater., 2021, 414, 125468.
48. H. Wang, J.-X. Liu, L. F. Allard, S. Lee, J. Liu, H. Li, J. Wang, J. Wang, S. H. Oh, W. Li, M. Flytzani-Stephanopoulos, M. Shen, B. R. Goldsmith and M. Yang, Nat. Commun., 2019, 10, 3808.
49. X. Tang, A. Yu, Q. Yang, H. Yuan, Z. Wang, J. Xie, L. Zhou, Y. Guo, D. Ma and S. Dai, J. Am. Chem. Soc., 2024, 146, 3764–3772.
50. M. Yoo, E. Kang, H. Choi, H. Ha, H. Choi, J.-S. Choi, K.-S. Lee, R. Celestre, D. A. Shapiro, J. Y. Park, C. Kim, Y.-S. Yu and H. Y. Kim, J. Mater. Chem. A, 2022, 10, 5942–5952.
51. J. Dong, D. Li, Y. Zhang, P. Chang and Q. Jin, J. Catal., 2022, 407, 174–185.
52. A. Aitbekova, C. Zhou, M. L. Stone, J. S. Lezama-Pacheco, A.-C. Yang, A. S. Hoffman, E. D. Goodman, P. Huber, J. F. Stebbins, K. C. Bustillo, P. Ercius, J. Ciston, S. R. Bare, P. N. Plessow and M. Cargnello, Nat. Mater., 2022, 21, 1290–1297.
53. J. Dong, Y. Zhang, H. Zou, P. Chang and Y. Guo, Catal. Sci. Technol., 2021, 11, 3047–3058.
54. M. J. Hülsey, B. Zhang, Z. Ma, H. Asakura, D. A. Do, W. Chen, T. Tanaka, P. Zhang, Z. Wu and N. Yan, Nat. Commun., 2019, 10, 1330.
55. M. Kottwitz, Y. Li, R. M. Palomino, Z. Liu, G. Wang, Q. Wu, J. Huang, J. Timoshenko, S. D. Senanayake, M. Balasubramanian, D. Lu, R. G. Nuzzo and A. I. Frenkel, ACS Catal., 2019, 9, 8738–8748.
56. R. Mi, D. Li, Z. Hu and R. T. Yang, ACS Catal., 2021, 11, 7876–7889.
57. X. I. Pereira-Hernández, A. DeLaRiva, V. Muravev, D. Kunwar, H. Xiong, B. Sudduth, M. Engelhard, L. Kovarik, E. J. M. Hensen, Y. Wang and A. K. Datye, Nat. Commun., 2019, 10, 1358.
58. L. Zhang, G. Spezzati, V. Muravev, M. A. Verheijen, B. Zijlstra, I. A. W. Filot, Y.-Q. Su, M.-W. Chang and E. J. M. Hensen, ACS Catal., 2021, 11, 5614–5627.
59. L. Nie, D. Mei, H. Xiong, B. Peng, Z. Ren, X. I. P. Hernandez, A. DeLaRiva, M. Wang, M. H. Engelhard, L. Kovarik, A. K. Datye and Y. Wang, Science, 2017, 358, 1419–1423.
60. G. Xu, C. Cai and T. Wang, J. Am. Chem. Soc., 2022, 144, 23089–23095.
61. F. Liu, R. Gao, C. Shi, L. Pan, Z.-F. Huang, X. Zhang and J.-J. Zou, J. Am. Chem. Soc., 2023, 145, 25252–25263.
62. N. Wang, Y. Wang, K. Li and Z. Wu, Mol. Catal., 2024, 552, 113669.
63. J. Tang, K. Jia, R. Zhang, C. Liu, X. Lin, T. Ge, X. Liu, Q. Zhao, W. Liu, D. Ma, H. Fan and J. Huang, ACS Catal., 2024, 14, 2463–2472.
64. J. Dong, Y. Zhang, D. Li, A. Adogwa, S. Huang, M. Yang, J. Yang and Q. Jin, Appl. Catal., B, 2023, 330, 122662.
65. M. E. Grass, Y. Zhang, D. R. Butcher, J. Y. Park, Y. Li, H. Bluhm, K. M. Bratlie, T. Zhang and G. A. Somorjai, Angew. Chem., Int. Ed., 2008, 47, 8893–8896.
66. H. Tanaka, K. Orita, A. Maede, H. Ishikawa, M. Miura, S. Arai, T. Higuchi, S. Ohta and S. Muto, Appl. Catal., A, 2021, 626, 118334.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4nj04248d

---